1. Variation in Mycenaean Greek Jeroen Vis University of Amsterdam / ACLC

2. Introduction Risch (1966): Discussion of variation in Mycenaean Greek (Pylos and Cnossos) with respect to the features –*mn > ma ~ mo –dat. s.: -ei ~ -i –word internal e ~ i Mycénien normal vs. Mycénien spécial Extended to Mycenae by Hajnal (1997) Variation of type du-wo ~ dwo, ka-ki-ja ~ ka-za ascribed to orthography but no detailed discussion of possible other parameters

3. Aims To discuss possible parameters that could lead to variation of this type in Mycenaean Greek –Orthography –Siever’s law –Morphological / phonological context –Dialectal differentiation –Dynamic property of the grammar itself To formalize the attested alternations To explore the contribution of phonolgical theory to Mycenaeology and v.v.

4. Structure Overview of the language and its orthographic system Discussion of the nature of alternations The formal analysis of variation

5. The language Mycenaean is a Greek language spoken in the southern Balkans. Its sources consist of ±7000 inscriptions on clay tablets, which are dated at about 1250 b.C.

6. Orthography Syllabograms of the type (C)V (Linear B) Conventional transcription in Latin characters r = l Laryngeal features are not distinct in orthography The onset of the syllable is written in full: pe-re = pre Syllable codas are omitted: a-to = artos

7. Alternations Hiatus resolution by means of anaptyxis of [j] and segmental fusion: /k h alkios/ > [k h alkijos] ~ [k h alk j os]‘bronze’ /Ci 1 V 2 C/ > [Ci 1 jV 2 C] ~ [C j 1 V 2 C] Hiatus resolution by means of anaptyxis of [υ] and consonant formation: /perusinua/ > [perusinuυa] ~ [perusinυa]‘yesterday’s nom. pl. /Cu 1 V 2 C/ > [Cu 1 υV 2 C] ~ [Cυ 1 V 2 CΣ]

8. Possible explanations: Orthography Siever’s law Morphological / phonological context Dialectal differentiation Dynamic property of the grammar itself

9. Orthography? -----> NO, because: Variation is very frequent and regular Variation is always of the same type Variation occurs systematically only in this phonological context The same type of variation occurs in other languages, e.g. Armenian, in which it is ascribed to phonological processes (Vaux 1998)

10. Siever’s law? -----> NO, because: Heavy syllable + *ij / *uw, light syllable + *j / *w, e.g. [kno:ssijos] (from Knossos) vs. [ar j oha] (better nom. pl. ) But: [υorg j on] ‘cultivate part.’ [sp h e:nυenta] ‘with wedges nom. pl. ’ [korijandna] ‘coriander’ [marat h uυon] ‘fennel’

11. Variation occurs also with respect to the very same word: [meletrija] ~ [meletr j a] ‘miller fem ’ [k h alkijos] ~ [k h alk j os] ‘bronze’ [laurant h ija] ~ [laurant hj a] ‘placename’ [hikk w ija] ~ [hikk j a] ‘charriot’ [perusinuυa] ~ [perusinυa] ‘yesterday’s nom. pl. ’ [duυo:] ~ [dυo:] ‘two’ [barakuυei] ~ [barakυei] ‘smargd’ [enualijos] ~ [enυalijos] ‘proper name’

12. Morphological / phonological context? -----> Partially? Preferred realization of certain affixes, e.g.: [-υent-], [-ijos], [-ija] No occurrence of labial + [υ] (OCP) -----> But not crucial: The above affixes may surface in both realizations Variation occurs also with respect to the very same word

13. Dialectal differentiation: Concept-based corpus linguistics: Measuring convergence and divergence in a concept-based way, an approach for charting distances between language varieties (Geeraerts 2004) Concept: hiatus solution of /Ci 1 V 2 C/ and /Cu 1 V 2 C/ Corpora: All attested complete words on the inscriptions of Pylos (Corpus 1) and Cnossos (Corpus 2)

14. Formula:

16. Concluding, no language external parameter (geography, orthography) or structural parameter (Siever’s law, morpho-phonological context) can adequately account for the attested variation As a result, the variation should be considered as a dynamic property of the language itself

17. Formalizing variation OT approaches: Unranked constraints (Anttila 1997) Parallel grammars (Tzakosta 2004) Stochastic constraints (Boersma & Hayes 2001)

18. Unranked constraints: Two (or more) constraints are unranked with respect to each other. During the evaluation of the hierarchy, a violation mark is assigned to both constraints, resulting in surface variation.

20. However, this would predict a distribution of 50% - 50% of both surface forms, which is not the case in Mycenaean Greek: j-anap. vs. fusionυ-anap. vs. υ-form. 80% vs. 20%20% vs. 80% Solution: adding more constraints (up to 3 per context) But then we would need 10 constraints to account for the distribution of variation, whereas the phenomenon itself can be analyzed by means of 4 What in the case of 99% vs. 1%?

21. Parallel grammars: Two different, but fully ranked grammars are activated by the speaker: Grammar A: Integrity >> Uniformity /k h alkia/ > [k h alk j a] Grammar B: Uniformity >> Integrity /k h alkia/ > [k h alkija] No account of the distribution of variation

22. Stochastic constraints: Constraints have an evaluation range instead of an evaluation point Evaluation ranges can overlap partially, thus resulting in surface variation: C1 C2

23. Uniformity Integrity Integrity Identity IO [son] 60 40 60 k h alkija (80%) vs. k h alk j a (20%) dυo: (80%) vs. duυo: (20%)

24. Conclusions The above discussed type of variation -uυV- ~ -υV- / -ijV- ~ - j V- cannot be ascribed to orthography, morphophonological parameters, Siever’s law or geographic divergence As a result, it should be interpreted as a dynamic property of the language itself Stochastic OT can adequately account for the variation and its unequal distribution

25. Contribution of linguistic theory to the interpretation of Mycenaean Greek and v.v.: –Concept based corpus-linguistics can define in more detail the nature of variation with respect to geographic distribution –Mycenaean data provide arguments in favour of stochastic OT as a tool of formalizing variation

26. Thank You

27. Selected references Tzakosta, M. (2004). Multiple parallel grammars in the acquisition of stress in Greek L1, diss. Universiteit Leiden Boersma, P. & B. Hayes (2001). “Empirical tests in the gradual learning algorithm” Linguistic Inquiry 32: 45-86 Vaux, B. (1998). The phonology of Armenian, Oxford: Clarendon press Geeraerts, D. (2004). Corpus-based sociolectometry, Classnotes at LOT Summerschool, Universiteit Utrecht Anttila, A. (1997). “Deriving variation from grammar: a study of Finnish genitives” Hinskens, F. et al. (eds.) Variation, change and phonological theory, Amsterdam: John Benjamins, 35-86 Risch, E. (1966). “Les différences dialectales dans le Mycénien” Palmer, L. & J. Chadwick (eds.) Proceedings of the Cambridge colloquium on Mycenaean studies, Cambridge: University press, 150-157 Hajnal, I. (1997). Sprachgeschichte des mykenischen Griechisch. Zur Frage der Differenzierung zwischen ‘Mycénien normal’ und ‘Mycénien spécial, suppl. ad Minos 14